A single close-up of a dog reacting, built so the eyes carry the whole shot.
A golden retriever sitting on a pale carpet in a bright living room, head tilted, ears lifting as it hears something off-frame to the left. Duration: 5 seconds. Aspect ratio 9:16. Camera: locked off at the dog's eye level, 50mm, f/2.0. No movement. The dog fills the middle third of the frame with headroom above. Action: the head tilt begins one beat into the shot, not at the start. Ears lift, then the eyes widen slightly. The tail is out of frame — the reaction is carried entirely by the head. Light: large soft window light from camera left, slightly above. Gentle fill from the right so the shadow side of the face keeps detail. Warm afternoon colour, no harsh highlights on the fur. Background: softly out of focus, suggesting a lived-in room — a sofa edge, a lamp. Nothing identifiable, nothing distracting. Negative: no cuts, no zoom, no text, no collar tags catching light, no human in frame.
The one-beat delay before the head tilt is what makes this read as a reaction rather than as a pose. Starting mid-action is the most common reason generated pet clips look artificial.
Swap the breed and the room, keep the structure.